A post-mortem is what teams write after a project fails. By then the lessons are too expensive. A pre-mortem is the same exercise, run before the project starts. The team imagines the failure first; the failure becomes a thing to design against rather than a thing to discover.

Pair this with the kill-criteria template. The pre-mortem surfaces the failure modes; the kill-criteria pre-commits the response. Both run before any code or copy ships.

Setup

Schedule: 60 minutes, before the project kicks off, with everyone who will be on it.
Materials: a quiet room, one whiteboard or shared doc, no laptops.
Pre-read for participants: a one-page brief on what the project is, what success looks like, and what the timeline is.

The opening prompt

Read aloud at the start of the session. Verbatim. The exact framing matters because the brain processes "we failed" differently from "we might fail".

It is twelve months from today. We launched the project we are about to start. It has failed. Not partially failed. Failed completely, in a way the company will reference as a cautionary tale. Take five minutes, by yourself, and write down every reason it could have failed.

The five-minute solo write

Each person writes their own list, silent, no discussion. The point is to surface the failure modes each person privately suspects but would not raise in a normal kickoff. Optimism is the default mode; the pre-mortem creates a structured permission to be pessimistic for five minutes.

Most lists run to ten or fifteen items. Anything fewer suggests the participant is still in optimism mode.

The round-robin share

Each person reads their list aloud. No discussion, no debate, no "yes but". Just the items, going around the room. The facilitator writes everything down on the whiteboard.

Items will repeat. Repetition is signal — when three people independently name the same failure mode, that is the failure mode the team is most likely to walk into. Cluster the repeats.

The clustering pass

After every list has been read, group the items by category. Most lists collapse into 5-7 buckets:

Distribution failure: we built the right thing, no one knew about it.
Adoption failure: we shipped it, customers did not stay.
Operational failure: the team broke down, the schedule slipped, the cost ballooned.
Market failure: the demand we assumed was not there.
Technical failure: the build was harder than the plan accounted for.
Team failure: a key person left, the cofounder relationship cracked, morale collapsed.
External failure: a competitor moved, regulation shifted, the market turned.

The top-three vote

Each person votes for the three failure modes they think are most likely. The top three are the ones the project plans against. Not all of them. Three.

Trying to defend against ten failure modes simultaneously means defending against none of them well. Three is the discipline.

The plan

For each of the top three, write down:

Failure mode: ______
Leading indicator we would see (in weeks 1-4) if this is happening: ______
Pre-committed action if the indicator fires: ______
Owner for monitoring: ______

The leading indicator is the part most teams skip. Without it, the failure becomes visible at week twelve, when the cost of recovery is the cost of the whole project. With it, the failure becomes visible at week three.

When to re-run

Once at the start of every initiative bigger than two weeks. Once at any major scope change. The cost of running it is sixty minutes; the cost of skipping it is the difference between a $3,000 project failure and a $30,000 one.

A note on the mood

Pre-mortems can feel grim if the facilitator lets them. The trick is to frame them as a forcing function, not a forecast. The team is not predicting failure; the team is identifying the few things worth defending against. Done well, the room finishes the session more confident than when it started, because the project plan now has structural defences the original brief did not.

The cost of a pre-mortem is sixty minutes. The cost of skipping it is whatever the project costs when it fails for a reason someone in the room would have named in week one.

Template from Marga Haus · margahaus.com/resources · Adapt and use freely. Attribution appreciated, not required.

Pre-mortem template